Set up an virtual env and load requirements.txt
To start weaviate use the docker-compose in the weaviate directory
docker-compose up
#Index Data
For reading the Confluence set:
CONFLUENCE_JSESSION_ID=
in .env file. The JSESSION_ID shall be get from the Browser cookies after logging in.
I tried to use the token, with no success.
In .env set values
AZURE_KEY=
AZURE_ENDPOINT=
AZURE_VERSION=
AZURE_EMBEDDING_DEPLOYMENT=
to be able to connect the embeddings in azure.
Make sure the weaviate Database is empty. Then:
set -o allexport; source .env; set +o allexport
cd reader
python ./reader.py
and wait.
#Run it
Weviate creates an index, you get it by:
http://127.0.0.1:8081/v1/schema
the index is named 'class'
Add the name to the value WEAVIATE_INDEX= in .env
Set the LLM Deplyoment to use to the value AZURE_LLM_DEPLOYMENT in `.env``
Load it by set -o allexport; source .env; set +o allexport in the console.
Start the frontend (backend is started in background then):
cd chat
python ./frontend.py
Visit: http://127.0.0.1:7860/ Type a question, get a result, type another question, get an error ;-)